#include <stdio.h>
#include <assert.h>

#include <cuda_runtime_api.h>

//#include "cuda-error.h"

const double MB = 1024*1024;
const double KB = 1024;

// ----------------------------------------
int main( int argc, char** argv )
{
	//try {
		int ndev;
		int err = cudaGetDeviceCount( &ndev );
		if (err != 0 && err != cudaErrorNoDevice) {
			fprintf( stderr, "cudaGetDeviceCount error %d: %s\n",
			         err, cudaGetErrorString(err) );
			return -1;
		}
		printf( "ndevices %d\n", ndev );
		for (int dev = 0; dev < ndev; ++dev) {
			struct cudaDeviceProp prop;
			err = cudaGetDeviceProperties( &prop, dev );
			if (err != 0) {
				fprintf( stderr, "cudaGetDeviceProperties( prop, %d ) error %d: %s\n",
				         dev, err, cudaGetErrorString(err) );
				continue;
			}
			printf( "device %d\n", dev );
			printf( "  name                          = %s\n",               prop.name                        );
			printf( "  asyncEngineCount              = %d\n",               prop.asyncEngineCount            );
			printf( "  canMapHostMemory              = %d\n",               prop.canMapHostMemory            );
			printf( "  capability major.minor        = %d.%d\n",            prop.major, prop.minor           );
			printf( "  clockRate                     = %.1f MHz\n",         prop.clockRate / 1000.           );
			printf( "  computeMode                   = %d\n",               prop.computeMode                 );
			printf( "  concurrentKernels             = %d\n",               prop.concurrentKernels           );
			printf( "  deviceOverlap                 = %d\n",               prop.deviceOverlap               );
			printf( "  ECCEnabled                    = %d\n",               prop.ECCEnabled                  );
			printf( "  integrated                    = %d\n",               prop.integrated                  );
			printf( "  kernelExecTimeoutEnabled      = %d\n",               prop.kernelExecTimeoutEnabled    );
			printf( "  l2CacheSize                   = %.1f KB\n",          prop.l2CacheSize / KB            );
			printf( "  maxGridSize                   = %5d x %5d x %5d\n",  prop.maxGridSize[0],         prop.maxGridSize[1],         prop.maxGridSize[2] );
			printf( "  maxTexture1D                  = %d\n",               prop.maxTexture1D                );
			printf( "  maxTexture1DLayered           = %5d x %5d\n",        prop.maxTexture1DLayered[0], prop.maxTexture1DLayered[1] );
			printf( "  maxTexture2D                  = %5d x %5d\n",        prop.maxTexture2D[0],        prop.maxTexture2D[1]        );
			printf( "  maxTexture2DLayered           = %5d x %5d x %5d\n",  prop.maxTexture2DLayered[0], prop.maxTexture2DLayered[1], prop.maxTexture2DLayered[2] );
			printf( "  maxTexture3D                  = %5d x %5d x %5d\n",  prop.maxTexture3D[0],        prop.maxTexture3D[1],        prop.maxTexture3D[2]        );
			printf( "  maxThreadsDim                 = %5d x %5d x %5d\n",  prop.maxThreadsDim[0],       prop.maxThreadsDim[1],       prop.maxThreadsDim[2]       );
			printf( "  maxThreadsPerBlock            = %d\n",               prop.maxThreadsPerBlock          );
			printf( "  maxThreadsPerMultiProcessor   = %d\n",               prop.maxThreadsPerMultiProcessor ); 
			printf( "  memoryBusWidth                = %d\n",               prop.memoryBusWidth              );
			printf( "  memoryClockRate               = %.1f MHz\n",         prop.memoryClockRate / 1000.     );
			printf( "  memPitch                      = %.1f MB\n",          prop.memPitch / MB               );
			printf( "  multiProcessorCount           = %d\n",               prop.multiProcessorCount         );
			printf( "  pciBusID                      = %d\n",               prop.pciBusID                    );
			printf( "  pciDeviceID                   = %d\n",               prop.pciDeviceID                 );
			printf( "  pciDomainID                   = %d\n",               prop.pciDomainID                 );
			printf( "  regsPerBlock                  = %d\n",               prop.regsPerBlock                );
			printf( "  sharedMemPerBlock             = %.1f KB\n",          prop.sharedMemPerBlock / KB      );
			printf( "  surfaceAlignment              = %ld\n",              prop.surfaceAlignment            );
			printf( "  tccDriver                     = %d\n",               prop.tccDriver                   );
			printf( "  textureAlignment              = %ld\n",              prop.textureAlignment            );
			printf( "  totalConstMem                 = %.1f KB\n",          prop.totalConstMem  / KB         );
			printf( "  totalGlobalMem                = %.1f MB\n",          prop.totalGlobalMem / MB         );
			printf( "  unifiedAddressing             = %d\n",               prop.unifiedAddressing           );
			printf( "  warpSize                      = %d\n",               prop.warpSize                    );
			printf( "\n" );
		}
	//}
	//catch( cudaError_t err ) {
	//	printf( "CUDA error: %s (%d)\n", cudaGetErrorString( err ), err );
	//}
	//catch( CUresult err ) {
	//	printf( "CUDA low-level error: %s (%d)\n", cuGetErrorString( err ), err );
	//}
	
	return 0;
}